A little further information - Device Net is really modified Can Bus. AB developed it to allow more than one master - Can Bus only allows 1 master. The protocol was then thrown open and is managed by ODVA (Open Device Net Vendors Association originally). Many people manufacture Device Net equipment - have a look at the list of manufacturers and equipment and you may be quite surprised. It is an open protocol.
I am not aware of the origins of Ethernet IP but once again it is managed by ODVA and the list of manufacturers is very extensive. Once again it is an open protocol.
CompoNet was developed by Omron and thrown open and is also an open protocol. There are a number of manufacturers that make CompoNet equipment but the main manufacturer is Omron. That to me is a great shame as while it can only have one master it is half the price of Device Net and does not suffer from having to have special cable, sensitive about power supplies and voltage drops, is much faster and is token ring based so does not suffer from data crashes or any of those nasty things. Runs on a standard twisted pair at a very rapid exchange rate (twisted pair over 30 metres 1000 I/O updates in about 1 millisecond - really fast) and can be configured in some quite complex ways, unlike Device Net. It also has explicit messaging capabilities but I am unaware of any company that has implemented it yet. It is my very favorite remote I/O network and is no fuss - just works really well.
To me Device Net is too expensive and fiddly, Ethernet IP suffers from the same problems as all Ethernet networks with crashing data unless managed switches are used, IT want to own it and make absolute pains in the 'A' of themselves. CompoNet rules for remote I/O in my book.